This paper examines the effects of a cueing hierarchy on naming in a patient with anomic aphasia. Using a single‐subject multiple baseline design across behaviors, the patient was trained to produce single inanimate nouns while generalization was tested to semantically related nouns matched for frequency of occurrence. Results showed successful acquisition and maintenance of trained words, but no generalization to untrained words. These data indicate that generalization does not occur as a natural by‐product of successful treatment and suggest, as pointed out by Baer, Wolf, and Risley , that “generalization should be programmed rather than expected or lamented”.